TDgpt

AI agent for advanced time-series analytics

Attend the Release Webinar

Highlights

Zero Learning Curve

Time-series forecasting, anomaly detection, imputation, and classification — in just a single SQL statement

Zero Initial Investment

Free, open-source software with general-purpose models and algorithms included out of the box

Zero Wasted Time

Update algorithms and models with TDgpt’s open SDK and see results in your apps — no development work required

Advanced Analytics in One SQL Statement

Time-Series Forecasting

Forecast future values from historical data with TDgpt’s FORECAST function:

FORECAST(column_expr, option_expr)
  option_expr:
  {"algo=expr1[,wncheck=1|0][,conf=conf_val][,every=every_val][,rows=rows_val][,start=start_ts_val][,expr2]"}

Example

LSTM forecasting with a model named lstm_foo, using input data with a periodicity of 10 samples and confidence interval of 95%, and white noise checking disabled:

SELECT _flow, _fhigh, _frowts, FORECAST(i32, "algo=lstm,model=lstm_foo,alpha=95,period=10,wncheck=0") FROM foo;

Anomaly Detection

Detect anomalies in your data with TDgpt’s anomaly window:

ANOMALY_WINDOW(column_name, option_expr)
  option_expr:
  {"algo=expr1 [,wncheck=1|0] [,expr2]"}

Example

Autoencoder-based detection with the trained model ad_autoencoder_foo:

SELECT COUNT(*), _WSTART FROM foo ANOMALY_WINDOW(col1, 'algo=autoencoder, model=ad_autoencoder_foo');

Use Cases

Forecasting Electricity Demand

Forecasting electricity demand helps optimize power generation and grid planning, improves operational efficiency, and reduces overall system costs. It enhances energy security by balancing supply and demand, and enables better decision-making in energy markets, leading to economic gains for both providers and consumers.

Demo

TDgpt analyzes historical power consumption data from the public UTSD dataset, with 30-minute interval readings, and forecasts electricity demand several hours into the future.

Try Now

Forecasting Wind Power Generation

Forecasting wind power generation is key to integrating renewable energy into the grid efficiently. It helps grid operators plan ahead, maintain stability, and reduce reliance on costly backup power. Accurate forecasts also enable better scheduling, minimize curtailment, and improve market participation for wind energy producers, ultimately lowering costs and increasing the reliability of clean energy supply.

Demo

TDgpt analyzes wind power generation data from a wind farm in the public UTSD dataset (collected at 4-second intervals) and forecasts power output every 15 minutes for the next 24 hours

Try Now

Anomaly Detection in IT Monitoring

Anomaly detection in IT monitoring enables early identification of unusual behavior across systems, helping teams detect issues like performance degradation, security breaches, or system failures before they escalate. By automatically flagging deviations from normal patterns, it reduces downtime, improves system reliability, and supports faster incident response, ultimately enhancing the overall efficiency and resilience of IT operations.

Demo

TDgpt detects anomalies in CPU usage data from a server in Amazon’s AWS East Coast data center during an API Gateway failure, part of the public NAB dataset. The gateway failure triggers repeated exception handling and retries, causing abnormal fluctuations in CPU usage.

Try Now

Request a Demo

See how TDgpt can transform data engineering for enterprises like yours